src/ignore.c

   1 /*
   2  * Copyright (C) the libgit2 contributors. All rights reserved.
   3  *
   4  * This file is part of libgit2, distributed under the GNU GPL v2 with
   5  * a Linking Exception. For full terms see the included COPYING file.
   6  */
   7
   8 #include "ignore.h"
   9
  10 #include "git2/ignore.h"
  11 #include "common.h"
  12 #include "attrcache.h"
  13 #include "path.h"
  14 #include "config.h"
  15 #include "wildmatch.h"
  16
  17 #define GIT_IGNORE_INTERNAL             "[internal]exclude"
  18
  19 #define GIT_IGNORE_DEFAULT_RULES ".\n..\n.git\n"
  20
  21 /**
  22  * A negative ignore pattern can negate a positive one without
  23  * wildcards if it is a basename only and equals the basename of
  24  * the positive pattern. Thus
  25  *
  26  * foo/bar
  27  * !bar
  28  *
  29  * would result in foo/bar being unignored again while
  30  *
  31  * moo/foo/bar
  32  * !foo/bar
  33  *
  34  * would do nothing. The reverse also holds true: a positive
  35  * basename pattern can be negated by unignoring the basename in
  36  * subdirectories. Thus
  37  *
  38  * bar
  39  * !foo/bar
  40  *
  41  * would result in foo/bar being unignored again. As with the
  42  * first case,
  43  *
  44  * foo/bar
  45  * !moo/foo/bar
  46  *
  47  * would do nothing, again.
  48  */
  49 static int does_negate_pattern(git_attr_fnmatch *rule, git_attr_fnmatch *neg)
  50 {
  51         int (*cmp)(const char *, const char *, size_t);
  52         git_attr_fnmatch *longer, *shorter;
  53         char *p;
  54
  55         if ((rule->flags & GIT_ATTR_FNMATCH_NEGATIVE) != 0
  56             || (neg->flags & GIT_ATTR_FNMATCH_NEGATIVE) == 0)
  57                 return false;
  58
  59         if (neg->flags & GIT_ATTR_FNMATCH_ICASE)
  60                 cmp = git__strncasecmp;
  61         else
  62                 cmp = git__strncmp;
  63
  64         /* If lengths match we need to have an exact match */
  65         if (rule->length == neg->length) {
  66                 return cmp(rule->pattern, neg->pattern, rule->length) == 0;
  67         } else if (rule->length < neg->length) {
  68                 shorter = rule;
  69                 longer = neg;
  70         } else {
  71                 shorter = neg;
  72                 longer = rule;
  73         }
  74
  75         /* Otherwise, we need to check if the shorter
  76          * rule is a basename only (that is, it contains
  77          * no path separator) and, if so, if it
  78          * matches the tail of the longer rule */
  79         p = longer->pattern + longer->length - shorter->length;
  80
  81         if (p[-1] != '/')
  82                 return false;
  83         if (memchr(shorter->pattern, '/', shorter->length) != NULL)
  84                 return false;
  85
  86         return cmp(p, shorter->pattern, shorter->length) == 0;
  87 }
  88
  89 /**
  90  * A negative ignore can only unignore a file which is given explicitly before, thus
  91  *
  92  *    foo
  93  *    !foo/bar
  94  *
  95  * does not unignore 'foo/bar' as it's not in the list. However
  96  *
  97  *    foo/<star>
  98  *    !foo/bar
  99  *
 100  * does unignore 'foo/bar', as it is contained within the 'foo/<star>' rule.
 101  */
 102 static int does_negate_rule(int *out, git_vector *rules, git_attr_fnmatch *match)
 103 {
 104         int error = 0, wildmatch_flags;
 105         size_t i;
 106         git_attr_fnmatch *rule;
 107         char *path;
 108         git_buf buf = GIT_BUF_INIT;
 109
 110         *out = 0;
 111
 112         wildmatch_flags = WM_PATHNAME;
 113         if (match->flags & GIT_ATTR_FNMATCH_ICASE)
 114                 wildmatch_flags |= WM_CASEFOLD;
 115
 116         /* path of the file relative to the workdir, so we match the rules in subdirs */
 117         if (match->containing_dir) {
 118                 git_buf_puts(&buf, match->containing_dir);
 119         }
 120         if (git_buf_puts(&buf, match->pattern) < 0)
 121                 return -1;
 122
 123         path = git_buf_detach(&buf);
 124
 125         git_vector_foreach(rules, i, rule) {
 126                 if (!(rule->flags & GIT_ATTR_FNMATCH_HASWILD)) {
 127                         if (does_negate_pattern(rule, match)) {
 128                                 error = 0;
 129                                 *out = 1;
 130                                 goto out;
 131                         }
 132                         else
 133                                 continue;
 134                 }
 135
 136                 git_buf_clear(&buf);
 137                 if (rule->containing_dir)
 138                         git_buf_puts(&buf, rule->containing_dir);
 139                 git_buf_puts(&buf, rule->pattern);
 140
 141                 if (git_buf_oom(&buf))
 142                         goto out;
 143
 144                 /* if we found a match, we want to keep this rule */
 145                 if ((wildmatch(git_buf_cstr(&buf), path, wildmatch_flags)) == WM_MATCH) {
 146                         *out = 1;
 147                         error = 0;
 148                         goto out;
 149                 }
 150         }
 151
 152         error = 0;
 153
 154 out:
 155         git__free(path);
 156         git_buf_dispose(&buf);
 157         return error;
 158 }
 159
 160 static int parse_ignore_file(
 161         git_repository *repo, git_attr_file *attrs, const char *data, bool allow_macros)
 162 {
 163         int error = 0;
 164         int ignore_case = false;
 165         const char *scan = data, *context = NULL;
 166         git_attr_fnmatch *match = NULL;
 167
 168         GIT_UNUSED(allow_macros);
 169
 170         if (git_repository__configmap_lookup(&ignore_case, repo, GIT_CONFIGMAP_IGNORECASE) < 0)
 171                 git_error_clear();
 172
 173         /* if subdir file path, convert context for file paths */
 174         if (attrs->entry &&
 175                 git_path_root(attrs->entry->path) < 0 &&
 176                 !git__suffixcmp(attrs->entry->path, "/" GIT_IGNORE_FILE))
 177                 context = attrs->entry->path;
 178
 179         if (git_mutex_lock(&attrs->lock) < 0) {
 180                 git_error_set(GIT_ERROR_OS, "failed to lock ignore file");
 181                 return -1;
 182         }
 183
 184         while (!error && *scan) {
 185                 int valid_rule = 1;
 186
 187                 if (!match && !(match = git__calloc(1, sizeof(*match)))) {
 188                         error = -1;
 189                         break;
 190                 }
 191
 192                 match->flags =
 193                     GIT_ATTR_FNMATCH_ALLOWSPACE | GIT_ATTR_FNMATCH_ALLOWNEG;
 194
 195                 if (!(error = git_attr_fnmatch__parse(
 196                         match, &attrs->pool, context, &scan)))
 197                 {
 198                         match->flags |= GIT_ATTR_FNMATCH_IGNORE;
 199
 200                         if (ignore_case)
 201                                 match->flags |= GIT_ATTR_FNMATCH_ICASE;
 202
 203                         scan = git__next_line(scan);
 204
 205                         /*
 206                          * If a negative match doesn't actually do anything,
 207                          * throw it away. As we cannot always verify whether a
 208                          * rule containing wildcards negates another rule, we
 209                          * do not optimize away these rules, though.
 210                          * */
 211                         if (match->flags & GIT_ATTR_FNMATCH_NEGATIVE
 212                             && !(match->flags & GIT_ATTR_FNMATCH_HASWILD))
 213                                 error = does_negate_rule(&valid_rule, &attrs->rules, match);
 214
 215                         if (!error && valid_rule)
 216                                 error = git_vector_insert(&attrs->rules, match);
 217                 }
 218
 219                 if (error != 0 || !valid_rule) {
 220                         match->pattern = NULL;
 221
 222                         if (error == GIT_ENOTFOUND)
 223                                 error = 0;
 224                 } else {
 225                         match = NULL; /* vector now "owns" the match */
 226                 }
 227         }
 228
 229         git_mutex_unlock(&attrs->lock);
 230         git__free(match);
 231
 232         return error;
 233 }
 234
 235 static int push_ignore_file(
 236         git_ignores *ignores,
 237         git_vector *which_list,
 238         const char *base,
 239         const char *filename)
 240 {
 241         int error = 0;
 242         git_attr_file *file = NULL;
 243
 244         error = git_attr_cache__get(&file, ignores->repo, NULL, GIT_ATTR_FILE__FROM_FILE,
 245                                     base, filename, parse_ignore_file, false);
 246         if (error < 0)
 247                 return error;
 248
 249         if (file != NULL) {
 250                 if ((error = git_vector_insert(which_list, file)) < 0)
 251                         git_attr_file__free(file);
 252         }
 253
 254         return error;
 255 }
 256
 257 static int push_one_ignore(void *payload, const char *path)
 258 {
 259         git_ignores *ign = payload;
 260         ign->depth++;
 261         return push_ignore_file(ign, &ign->ign_path, path, GIT_IGNORE_FILE);
 262 }
 263
 264 static int get_internal_ignores(git_attr_file **out, git_repository *repo)
 265 {
 266         int error;
 267
 268         if ((error = git_attr_cache__init(repo)) < 0)
 269                 return error;
 270
 271         error = git_attr_cache__get(out, repo, NULL, GIT_ATTR_FILE__IN_MEMORY, NULL,
 272                                     GIT_IGNORE_INTERNAL, NULL, false);
 273
 274         /* if internal rules list is empty, insert default rules */
 275         if (!error && !(*out)->rules.length)
 276                 error = parse_ignore_file(repo, *out, GIT_IGNORE_DEFAULT_RULES, false);
 277
 278         return error;
 279 }
 280
 281 int git_ignore__for_path(
 282         git_repository *repo,
 283         const char *path,
 284         git_ignores *ignores)
 285 {
 286         int error = 0;
 287         const char *workdir = git_repository_workdir(repo);
 288         git_buf infopath = GIT_BUF_INIT;
 289
 290         assert(repo && ignores && path);
 291
 292         memset(ignores, 0, sizeof(*ignores));
 293         ignores->repo = repo;
 294
 295         /* Read the ignore_case flag */
 296         if ((error = git_repository__configmap_lookup(
 297                         &ignores->ignore_case, repo, GIT_CONFIGMAP_IGNORECASE)) < 0)
 298                 goto cleanup;
 299
 300         if ((error = git_attr_cache__init(repo)) < 0)
 301                 goto cleanup;
 302
 303         /* given a unrooted path in a non-bare repo, resolve it */
 304         if (workdir && git_path_root(path) < 0) {
 305                 git_buf local = GIT_BUF_INIT;
 306
 307                 if ((error = git_path_dirname_r(&local, path)) < 0 ||
 308                     (error = git_path_resolve_relative(&local, 0)) < 0 ||
 309                     (error = git_path_to_dir(&local)) < 0 ||
 310                     (error = git_buf_joinpath(&ignores->dir, workdir, local.ptr)) < 0)
 311                 {;} /* Nothing, we just want to stop on the first error */
 312                 git_buf_dispose(&local);
 313         } else {
 314                 error = git_buf_joinpath(&ignores->dir, path, "");
 315         }
 316         if (error < 0)
 317                 goto cleanup;
 318
 319         if (workdir && !git__prefixcmp(ignores->dir.ptr, workdir))
 320                 ignores->dir_root = strlen(workdir);
 321
 322         /* set up internals */
 323         if ((error = get_internal_ignores(&ignores->ign_internal, repo)) < 0)
 324                 goto cleanup;
 325
 326         /* load .gitignore up the path */
 327         if (workdir != NULL) {
 328                 error = git_path_walk_up(
 329                         &ignores->dir, workdir, push_one_ignore, ignores);
 330                 if (error < 0)
 331                         goto cleanup;
 332         }
 333
 334         /* load .git/info/exclude if possible */
 335         if ((error = git_repository_item_path(&infopath, repo, GIT_REPOSITORY_ITEM_INFO)) < 0 ||
 336                 (error = push_ignore_file(ignores, &ignores->ign_global, infopath.ptr, GIT_IGNORE_FILE_INREPO)) < 0) {
 337                 if (error != GIT_ENOTFOUND)
 338                         goto cleanup;
 339                 error = 0;
 340         }
 341
 342         /* load core.excludesfile */
 343         if (git_repository_attr_cache(repo)->cfg_excl_file != NULL)
 344                 error = push_ignore_file(
 345                         ignores, &ignores->ign_global, NULL,
 346                         git_repository_attr_cache(repo)->cfg_excl_file);
 347
 348 cleanup:
 349         git_buf_dispose(&infopath);
 350         if (error < 0)
 351                 git_ignore__free(ignores);
 352
 353         return error;
 354 }
 355
 356 int git_ignore__push_dir(git_ignores *ign, const char *dir)
 357 {
 358         if (git_buf_joinpath(&ign->dir, ign->dir.ptr, dir) < 0)
 359                 return -1;
 360
 361         ign->depth++;
 362
 363         return push_ignore_file(
 364                 ign, &ign->ign_path, ign->dir.ptr, GIT_IGNORE_FILE);
 365 }
 366
 367 int git_ignore__pop_dir(git_ignores *ign)
 368 {
 369         if (ign->ign_path.length > 0) {
 370                 git_attr_file *file = git_vector_last(&ign->ign_path);
 371                 const char *start = file->entry->path, *end;
 372
 373                 /* - ign->dir looks something like "/home/user/a/b/" (or "a/b/c/d/")
 374                  * - file->path looks something like "a/b/.gitignore
 375                  *
 376                  * We are popping the last directory off ign->dir.  We also want
 377                  * to remove the file from the vector if the popped directory
 378                  * matches the ignore path.  We need to test if the "a/b" part of
 379                  * the file key matches the path we are about to pop.
 380                  */
 381
 382                 if ((end = strrchr(start, '/')) != NULL) {
 383                         size_t dirlen = (end - start) + 1;
 384                         const char *relpath = ign->dir.ptr + ign->dir_root;
 385                         size_t pathlen = ign->dir.size - ign->dir_root;
 386
 387                         if (pathlen == dirlen && !memcmp(relpath, start, dirlen)) {
 388                                 git_vector_pop(&ign->ign_path);
 389                                 git_attr_file__free(file);
 390                         }
 391                 }
 392         }
 393
 394         if (--ign->depth > 0) {
 395                 git_buf_rtruncate_at_char(&ign->dir, '/');
 396                 git_path_to_dir(&ign->dir);
 397         }
 398
 399         return 0;
 400 }
 401
 402 void git_ignore__free(git_ignores *ignores)
 403 {
 404         unsigned int i;
 405         git_attr_file *file;
 406
 407         git_attr_file__free(ignores->ign_internal);
 408
 409         git_vector_foreach(&ignores->ign_path, i, file) {
 410                 git_attr_file__free(file);
 411                 ignores->ign_path.contents[i] = NULL;
 412         }
 413         git_vector_free(&ignores->ign_path);
 414
 415         git_vector_foreach(&ignores->ign_global, i, file) {
 416                 git_attr_file__free(file);
 417                 ignores->ign_global.contents[i] = NULL;
 418         }
 419         git_vector_free(&ignores->ign_global);
 420
 421         git_buf_dispose(&ignores->dir);
 422 }
 423
 424 static bool ignore_lookup_in_rules(
 425         int *ignored, git_attr_file *file, git_attr_path *path)
 426 {
 427         size_t j;
 428         git_attr_fnmatch *match;
 429
 430         git_vector_rforeach(&file->rules, j, match) {
 431                 if (match->flags & GIT_ATTR_FNMATCH_DIRECTORY &&
 432                     path->is_dir == GIT_DIR_FLAG_FALSE)
 433                         continue;
 434                 if (git_attr_fnmatch__match(match, path)) {
 435                         *ignored = ((match->flags & GIT_ATTR_FNMATCH_NEGATIVE) == 0) ?
 436                                 GIT_IGNORE_TRUE : GIT_IGNORE_FALSE;
 437                         return true;
 438                 }
 439         }
 440
 441         return false;
 442 }
 443
 444 int git_ignore__lookup(
 445         int *out, git_ignores *ignores, const char *pathname, git_dir_flag dir_flag)
 446 {
 447         size_t i;
 448         git_attr_file *file;
 449         git_attr_path path;
 450
 451         *out = GIT_IGNORE_NOTFOUND;
 452
 453         if (git_attr_path__init(
 454                 &path, pathname, git_repository_workdir(ignores->repo), dir_flag) < 0)
 455                 return -1;
 456
 457         /* first process builtins - success means path was found */
 458         if (ignore_lookup_in_rules(out, ignores->ign_internal, &path))
 459                 goto cleanup;
 460
 461         /* next process files in the path.
 462          * this process has to process ignores in reverse order
 463          * to ensure correct prioritization of rules
 464          */
 465         git_vector_rforeach(&ignores->ign_path, i, file) {
 466                 if (ignore_lookup_in_rules(out, file, &path))
 467                         goto cleanup;
 468         }
 469
 470         /* last process global ignores */
 471         git_vector_foreach(&ignores->ign_global, i, file) {
 472                 if (ignore_lookup_in_rules(out, file, &path))
 473                         goto cleanup;
 474         }
 475
 476 cleanup:
 477         git_attr_path__free(&path);
 478         return 0;
 479 }
 480
 481 int git_ignore_add_rule(git_repository *repo, const char *rules)
 482 {
 483         int error;
 484         git_attr_file *ign_internal = NULL;
 485
 486         if ((error = get_internal_ignores(&ign_internal, repo)) < 0)
 487                 return error;
 488
 489         error = parse_ignore_file(repo, ign_internal, rules, false);
 490         git_attr_file__free(ign_internal);
 491
 492         return error;
 493 }
 494
 495 int git_ignore_clear_internal_rules(git_repository *repo)
 496 {
 497         int error;
 498         git_attr_file *ign_internal;
 499
 500         if ((error = get_internal_ignores(&ign_internal, repo)) < 0)
 501                 return error;
 502
 503         if (!(error = git_attr_file__clear_rules(ign_internal, true)))
 504                 error = parse_ignore_file(
 505                                 repo, ign_internal, GIT_IGNORE_DEFAULT_RULES, false);
 506
 507         git_attr_file__free(ign_internal);
 508         return error;
 509 }
 510
 511 int git_ignore_path_is_ignored(
 512         int *ignored,
 513         git_repository *repo,
 514         const char *pathname)
 515 {
 516         int error;
 517         const char *workdir;
 518         git_attr_path path;
 519         git_ignores ignores;
 520         unsigned int i;
 521         git_attr_file *file;
 522         git_dir_flag dir_flag = GIT_DIR_FLAG_UNKNOWN;
 523
 524         assert(repo && ignored && pathname);
 525
 526         workdir = git_repository_workdir(repo);
 527
 528         memset(&path, 0, sizeof(path));
 529         memset(&ignores, 0, sizeof(ignores));
 530
 531         if (!git__suffixcmp(pathname, "/"))
 532                 dir_flag = GIT_DIR_FLAG_TRUE;
 533         else if (git_repository_is_bare(repo))
 534                 dir_flag = GIT_DIR_FLAG_FALSE;
 535
 536         if ((error = git_attr_path__init(&path, pathname, workdir, dir_flag)) < 0 ||
 537                 (error = git_ignore__for_path(repo, path.path, &ignores)) < 0)
 538                 goto cleanup;
 539
 540         while (1) {
 541                 /* first process builtins - success means path was found */
 542                 if (ignore_lookup_in_rules(ignored, ignores.ign_internal, &path))
 543                         goto cleanup;
 544
 545                 /* next process files in the path */
 546                 git_vector_foreach(&ignores.ign_path, i, file) {
 547                         if (ignore_lookup_in_rules(ignored, file, &path))
 548                                 goto cleanup;
 549                 }
 550
 551                 /* last process global ignores */
 552                 git_vector_foreach(&ignores.ign_global, i, file) {
 553                         if (ignore_lookup_in_rules(ignored, file, &path))
 554                                 goto cleanup;
 555                 }
 556
 557                 /* move up one directory */
 558                 if (path.basename == path.path)
 559                         break;
 560                 path.basename[-1] = '\0';
 561                 while (path.basename > path.path && *path.basename != '/')
 562                         path.basename--;
 563                 if (path.basename > path.path)
 564                         path.basename++;
 565                 path.is_dir = 1;
 566
 567                 if ((error = git_ignore__pop_dir(&ignores)) < 0)
 568                         break;
 569         }
 570
 571         *ignored = 0;
 572
 573 cleanup:
 574         git_attr_path__free(&path);
 575         git_ignore__free(&ignores);
 576         return error;
 577 }
 578
 579 int git_ignore__check_pathspec_for_exact_ignores(
 580         git_repository *repo,
 581         git_vector *vspec,
 582         bool no_fnmatch)
 583 {
 584         int error = 0;
 585         size_t i;
 586         git_attr_fnmatch *match;
 587         int ignored;
 588         git_buf path = GIT_BUF_INIT;
 589         const char *wd, *filename;
 590         git_index *idx;
 591
 592         if ((error = git_repository__ensure_not_bare(
 593                         repo, "validate pathspec")) < 0 ||
 594                 (error = git_repository_index(&idx, repo)) < 0)
 595                 return error;
 596
 597         wd = git_repository_workdir(repo);
 598
 599         git_vector_foreach(vspec, i, match) {
 600                 /* skip wildcard matches (if they are being used) */
 601                 if ((match->flags & GIT_ATTR_FNMATCH_HASWILD) != 0 &&
 602                         !no_fnmatch)
 603                         continue;
 604
 605                 filename = match->pattern;
 606
 607                 /* if file is already in the index, it's fine */
 608                 if (git_index_get_bypath(idx, filename, 0) != NULL)
 609                         continue;
 610
 611                 if ((error = git_buf_joinpath(&path, wd, filename)) < 0)
 612                         break;
 613
 614                 /* is there a file on disk that matches this exactly? */
 615                 if (!git_path_isfile(path.ptr))
 616                         continue;
 617
 618                 /* is that file ignored? */
 619                 if ((error = git_ignore_path_is_ignored(&ignored, repo, filename)) < 0)
 620                         break;
 621
 622                 if (ignored) {
 623                         git_error_set(GIT_ERROR_INVALID, "pathspec contains ignored file '%s'",
 624                                 filename);
 625                         error = GIT_EINVALIDSPEC;
 626                         break;
 627                 }
 628         }
 629
 630         git_index_free(idx);
 631         git_buf_dispose(&path);
 632
 633         return error;
 634 }
 635